Projects

The wireless sensor networks and ad-hoc networks have been extensively researched over the past decade. Such networks have been employed for several applications such as environmental monitoring, military surveillance, and industrial data collection and reporting. Another promising application is a wireless body area network (BAN), which is a sensor and ad-hoc network around the human body. Wireless BANs have the potential to be widely utilized, with applications ranging from medical monitoring and rehabilitation to wearable computing and sensing in both the military and common human life. Although wireless BANs are an emerging technology that can improve human life in many aspects, the following issues regarding the wireless BAN communication channel are still considered as critical limitations:

The wireless sensor networks and ad-hoc networks have been extensively researched over the past decade. Such networks have been employed for several applications such as environmental monitoring, military surveillance, and industrial data collection and reporting. Another promising application is a wireless body area network (BAN), which is a sensor and ad-hoc network around the human body. Wireless BANs have the potential to be widely utilized, with applications ranging from medical monitoring and rehabilitation to wearable computing and sensing in both the military and common human life. Although wireless BANs are an emerging technology that can improve human life in many aspects, the following issues regarding the wireless BAN communication channel are still considered as critical limitations:

1) interference on a wireless communication channel caused by multiple sensors/devices;

2) the lack of system integration of individual sensors;

3) the necessity for long-term low power consumption.

This research proposes a theoretical model for the polarized BAN channel, characterized by either isotropic or non-isotropic azimuthal scattering and with either non-line-of-sight (NLoS) or line-of-sight (LoS) conditions between a Tx and Rx that are both located on the body. The main contributions of the research are summarized as follows:

1) Unique three-dimensional (3-D) geometry-based model for the polarized wireless BAN channel with three different propagation modes, i.e., cylindrical-surface-scattered (CSS), body-scattered (BS) and ground-scattered (GS) propagation modes;

2) Theoretical modeling of the creeping wave using diffraction theories;

3) Empirical validation of the proposed polarized BAN channel model;

4) Observation that the GS propagation mode is the most dominant propagation mode in the time-frequencycorrelation function (TF-CF) characteristics;

5) Analytical characterization of polarized BAN channels based on the proposed BAN channel model.

Wireless communication devices with multiple network interface cards (NICs) have been considered for exploiting the availability of multiple channels in wireless networks. For instance, such a network is standardized in IEEE~802.11a/h/j/n for 5 GHz carrier frequency band and IEEE~802.11b/g/n for 2.4 GHz band, depending on the regulatory region. In the wireless networking literature, communication devices having multiple network interface cards are called multi-interface nodes, and consist of multiple NICs in the physical sense or multiple half-duplex transceivers in the functional sense. In spite of such multi-channel/interface devices, traffic saturation can still occur in heavy traffic load areas such as the nodes at the center of large-scale wireless ad hoc networks.

Network coding can mitigate traffic saturation by increasing the aggregate network throughput as the network traffic increases. Since the original paper of Ahlswede et al., many researchers have considered the practical aspects of network coding. Lun et al. proposed a minimum-energy multicasting scheme for wireless networks where they showed that network coding can improve throughput in a multiple unicast scenario. A new architecture, COPE, was proposed for wireless mesh networks, which performs exclusive OR (XOR) operations on pairs of packets from different nodes in multiple unicasts. For COPE-type network coding, Li and Chiu presented a routing algorithm combined with opportunistic network coding. Moreover, they showed that radio coverage limitations will prevent the encoding of some packets that traverse an intersecting node. Prior literature presents coding-aware routing algorithms for XOR encoding at intersecting nodes that reduces network traffic.

While the aforementioned studies mainly considered single-channel networks, our research extends the concepts to multi-channel/interface scenarios. In contrast to traditional network coding, our network-coding scheme uses coded-overhearing to cope with radio coverage limitations. Our algorithm also distributes the load of the intersecting node to its neighboring nodes by dispersing the required number of interfaces around the intersecting node. Thus, our network coding scheme improves the performance when the number of interfaces at each node is limited. Although there are prior approaches that propose different sorts of channel assignment algorithms for multi-channel/interface wireless networks, they do not consider network coding gain.

This project describes a new channel assignment algorithm that accounts for network coding gain, available channel/interface capacity, and expected waiting time for network coding opportunities. Our main contributions are summarized as follows:

1) a novel concept that combines multi-channel/interface network coding with channel assignment;

2) a network coding scheme with coded-overhearing and a coding-aware channel assignment algorithm that are both novelties by themselves. Their combination is shown to provide a substantial improvement in the aggregate network throughput.